Ca2+-dependent exocytotic pathways in Chinese hamster ovary fibroblasts revealed by a caged-Ca2+ compound.

Ca2+-dependent exocytosis and endocytosis of Chinese hamster ovary (CHO) fibroblasts were investigated using capacitance measurement and rapid photolysis of a caged-Ca2+ compound, dimethoxynitrophenamine tetrasodium salt. CHO cells exhibited large and fast increases in membrane capacitance (1.9 ± 1 picofarads, or 13 ± 7% of total membrane area, mean ± S.D., n = 37) upon Ca2+ jumps to [Ca2+]i larger than 20 μM. The fast exocytosis occurred with a delay (20-80 ms), and exhibited a rate constant that was strongly dependent on [Ca2+]i. The maximal rate constant of exocytosis was 2.8/s, and a half-maximal rate was achieved at 30 μM. The fast exocytosis was followed by rapid endocytosis in 28% of the cells. The endocytosis often began after a delay of 0.5-2 s. Ca2+ jumps also induced stepwise increases in membrane capacitance of 10-134 femtofarads in 40% of the cells, indicating fusion of large vesicles with diameters of 0.4-1.5 μm. The exocytosis of the large vesicles could selectively be induced with smaller Ca2+ jumps (6-20 μM), and occurred slowly with a rate constant of 0.3/s. These data indicate that CHO fibroblasts possess Ca2+-dependent exocytotic mechanisms. Moreover, two parallel exocytotic pathways may exist reminiscent of those of neurons and endocrine cells. A kinetic model was constructed to account for the fast exocytosis of CHO cells.

Ca 2ϩ -dependent exocytotic secretion of neurotransmitters and hormones has been considered as a function representative of neurons and endocrine cells (1)(2)(3). It is well established that neurons possess two distinct types of secretory vesicles, the synaptic vesicles carrying classical neurotransmitters and large dense-core vesicles carrying neuropeptides. Some endocrine cells also possess synaptic-like microvesicles in addition to dense-core vesicles (4). The synaptic-like microvesicles are generated by recycling between the plasma membrane and early endosomes as in the case of the synaptic vesicles (1), and they express synaptophysin as a common marker. Recently, it has been reported that even non-secretory cells possess recycling membrane pathways, and transfected synaptophysin selectively distributes to these early endosomal vesicles (5,6). In addition, putative Ca 2ϩ sensor molecules of exocytosis, synaptotagmins, are ubiquitously expressed (7). Furthermore, Ca 2ϩdependent secretion of a fast neurotransmitter, ACh, has been reported in muscles and fibroblasts that were incubated with ACh (8,9).
In order to characterize Ca 2ϩ and time dependence of exocytosis of non-secretory cells, we investigated exocytosis of CHO fibroblasts using time-resolved capacitance measurement in combination with rapid photolysis of a caged-Ca 2ϩ compound (10 -12). We found that the CHO cells undergo massive increases in membrane capacitance upon large increases in [Ca 2ϩ ] i 1 and exhibited several characteristics similar to those found in one type of nerve terminal and endocrine cells (10 -13). The capacitance increases occurred with a rate constant of 2.8/s and exhibited low sensitivity to Ca 2ϩ . Some cells even showed rapid endocytosis as in endocrine cells. Interestingly, we detected stepwise capacitance increases of 10 fF or larger in 40% of the cells, suggesting fusion of vesicles with diameters of 0.4 m or greater. The latter form of exocytosis occurred slowly and could be selectively induced by a small increase in [Ca 2ϩ ] i (Ͼ6 M). Thus, the fibroblasts have been found to possess two exocytotic pathways, which might correspond to the small-clear and large dense-core vesicles in neurons and endocrine cells.

EXPERIMENTAL PROCEDURES
Preparation of Cells-CHO cells were grown in Dulbecco's modified Eagle's medium (Life Technologies, Inc.) supplemented with 10% fetal calf serum and 1% penicillin-streptomycin (Life Technologies, Inc.) in an atmosphere of 10% CO 2 at 37°C. They were passaged approximately once a week, and plated on 5-mm cover glasses (No. 0, Menzel Glass, Braunschweig, Germany) in 96-well culture plates 1 to 3 days before the patch clamp experiments.
Recording Solutions-The external recording solution contained 140 mM NaCl, 5 mM KCl, 2 mM CaCl 2 , 1 mM MgCl 2 , 10 mM Na-HEPES (Dojin, Kumamoto, Japan), and 10 mM glucose at pH 7.4 and 320 mosm. A caged-Ca 2ϩ compound and other chemicals were dissolved in a basic internal solution containing 120 mM Cs-glutamate, 10 mM CsCl, 40 mM Cs-HEPES, and 200 M benzothiazole coumarin (Molecular Probes, Eugene, OR) at pH 7.3. As a caged-Ca 2ϩ compound, we used 10 mM DM-nitrophen (dimethoxynitrophenamine tetrasodium salt; Calbiochem, La Jolla, CA) loaded with 4 mM CaCl 2 . The osmolarity of the internal solutions was about 320 mM after addition of these chemicals. Intermediate [Ca 2ϩ ] i levels were created by loading DM-nitrophen with lesser amounts of CaCl 2 (2.5, 2, 1 or 0.5 mM). We carried out all experimental procedures under yellow illumination (FL40S-Y-F, National, Tokyo) at room temperature, 22-25°C.
Capacitance Measurements-The capacitance measurements were carried out using a patch clamp amplifier, AxoPatch 1D (Axon Instrument, Foster City, CA), and the phase tracking method (12,14). A computer-based lock-in amplifier was constructed using the C programming language and a personal computer (NEC 9821An, Tokyo, Japan) (15). The cells were held at Ϫ20 mV, at which current-to-voltage relationships became almost linear soon after the whole cell recording and to which 1 kHz sine waves with a peak to peak amplitude of 100 mV were applied. Capacitance and conductance were calculated from 10 cycles of sine waves and stored at 83 Hz, or they were calculated from one cycle of sine wave and stored at 1 kHz for the experiments using a * This work was supported by grants-in-aid from the Japanese Ministry of Education, Science, and Cultures, a research grant from the Human Frontier Science Organization, and a grant from the Takeda Science Foundation. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
‡ To whom correspondence should be addressed: Dept. of Physiology, Faculty of Medicine, University of Tokyo, Hongo, Bunkyo-ku, Tokyo 113, Japan. Fax: 81-3-3811-5520; E-mail: hkasai@m.u-tokyo.ac.jp. flash lamp. The phase tracking was performed every 7 s. The Ca 2ϩ jumps often induce transient inward currents (10). In order to minimize the error in the estimates of membrane capacitance due to the current, we selected our data according to the following criteria: 1) resistance of seal before photolysis was larger than 1 G⍀; 2) access resistance was smaller than 8 M⍀; and 3) amplitude of the transient current was less than 100 pA (5 nS). Abiding by these criteria, the maximal error in the estimate of capacitance should be smaller than 0.16% of total membrane capacitance (16). Our studies were based on a total of 77 cells meeting these criteria. Whole cell membrane capacitance ranged between 8 and 30 pF (mean Ϯ S.D. ϭ 14.7 Ϯ 6.7 pF).
Ca 2ϩ Measurements-Measurement of [Ca 2ϩ ] i was performed using a ratioable long wavelength Ca 2ϩ -indicator dye, benzothiazole coumarin (17). Two monochromatic lights with wavelengths of 430 and 480 nm were isolated from a xenon lamp, rapidly alternated using a monochromator (T.I.L.L. Photonics, Munich, Germany) and fed into one port of a light guide (IX-RFA-caged, Olympus). The excitation light was reflected by a dichroic mirror, DM500, placed beneath the objective, and fluorescent light emitted from the cells was further filtered with an LP520 before entering a photomultiplier (NT5783, Hamamatsu Photonics, Hamamatsu, Japan) that was attached to the side port of the IX. One ratiometric Ca 2ϩ measurement was carried out during one capacitance measurement.
In vivo calibration experiments (18) of benzothiazole coumarin were performed in the presence of DM-nitrophen, because fluorescence properties of Ca 2ϩ indicators can be altered by DM-nitrophen (19). We prepared altogether 6 calibration solutions using the basic internal solution and either the caged or photolysed forms of 10 mM DM-nitrophen, whose Ca 2ϩ concentrations were adjusted to 0 M, 20 M, or 10 mM using Ca 2ϩ buffers and CaCl 2 . For 0 M and 10 mM Ca 2ϩ calibration solutions, we added 10 mM EGTA and 20 mM CaCl 2 , respectively, to the internal solutions with DM-nitrophen. To prepare the 20 M Ca 2ϩ solution, we utilized APTRA (2-aminophenol-N,N,O-triacetic acid, Molecular Probes) as a Ca 2ϩ buffer, since its K Ca is around 25 M. We first added 20 mM APTRA plus 8 mM CaCl 2 to the internal solution with DM-nitrophen, and titrated to 20 M with CaCl 2 using a mini Ca 2ϩ electrode, which was made of polyethylene tubing ( ϭ 1.5 mm) and a Ca 2ϩ ionophore, ETH 129 (Fluka, Switzerland) (20). A photolysed form of DM-nitrophen was generated by UV irradiation of DM-nitrophen for 20 min in a 96-culture well. These solutions were injected into CHO cells through patch pipettes, and fluorescence ratios were obtained a few minutes after loading when the ratios had stabilized. Calibration constants (R max , R min , and K d Ј) were obtained for each form of DMnitrophen as described previously (18). R max and R min were estimated as 2.5 and 0.4, respectively, for both the caged and photolysed DMnitrophen. Values of K d Ј were 102 and 141 M for caged and photolysed DM-nitrophen, respectively. The value for the photolysed DM-nitrophen was utilized to estimate the peak values of [Ca 2ϩ ] i , since most DM-nitrophen was photolysed upon UV irradiation in our experiments (see below). The error in the estimates of [Ca 2ϩ ] i during re-equilibration of caged Ca 2ϩ compounds after photolysis was calculated to be less than 16%.
Photolysis of Caged-Ca 2ϩ Compounds-We chiefly presented those data obtained using a mercury lamp (IX-RFC, Olympus) as an actinic light source; however, a xenon-flash lamp (High-Tech Instrument, United Kingdom) was utilized in some recent experiments. The light from the mercury lamp was filtered through a 360-nm band pass filter, and fed into the second port of the light guide (IX-RFA caged, Olympus). Equipped with a dichroic mirror, DM400, the light guide can accommodate two light sources, one for the actinic light and the other for excitation of a Ca 2ϩ -indicator dye. Irradiation with the actinic light was gated by an electric shutter (Copal, Tokyo). The duration of the opening of the shutter was set at 33 ms, which was more than enough to fully activate the caged compounds within the cells. The time constant of the activation of the caged compounds was estimated as 12 ms using the line scanning mode of a confocal microscope (MRC600, Bio-Rad) (21) and a droplet containing 200 M fluo-3, 100 M DM-nitrophen, and 50 M CaCl 2 . The speed of photolysis should not much affect the rate of the fast exocytosis, since the maximal rate constant of the fast exocytosis was below 3/s. In fact, the kinetics of the fast exocytosis was not affected by the use of the xenon flash lamp as a source of actinic light (n ϭ 9), with which the photolysis could be carried out within a fraction of a millisecond (13).

RESULTS
The fast Ca 2ϩ -dependent Exocytosis-Most CHO cells exhibited large increases in membrane capacitance when concentra-tions of intracellular Ca 2ϩ ([Ca 2ϩ ] i ) were rapidly raised to more than 20 M (Fig. 1, A, B, and D). The time course of the capacitance increase could be well described by a single exponential function with a time constant of 0.4 -2 s (Fig. 1B). Additional slow components seldom appeared even at greater [Ca 2ϩ ] i , unlike in the case of some endocrine cells (10 -12). The time course of the capacitance increases can be interpreted as the fusion and exhaustion of a release-ready pool of secretory vesicles, because little capacitance increase was evoked by the second Ca 2ϩ jump (data not shown). The rate constant of the fusion of vesicles was then estimated to be the inverse of the time constant. Fig. 1E shows a plot of the Ca 2ϩ dependence of the rate constant determined for 77 cells. The rate constant showed strong Ca 2ϩ dependence, was half-maximal at 30 M, and reached the maximal value of 2.8/s at [Ca 2ϩ ] i greater than 80 M. The Hill coefficient of the Ca 2ϩ dependence was 3.5 (a dashed line in Fig. 1E). Stepwise increases in capacitance reflecting single fusion events could not be resolved for the larger part of the capacitance increase ( Fig. 1B; but see below), suggesting that the exocytosis was mediated by small vesicles. We will refer to this fast and smooth capacitance change as fast exocytosis. The fast exocytosis, once evoked, amounted to 1.9 Ϯ 1 pF, or 13 Ϯ 7% (mean Ϯ S.E., n ϭ 37) of the total membrane area, within 5 s (Fig. 1D), and was not detected at [Ca 2ϩ ] i smaller than 10 M (Fig. 1, A and D).
Delays in the fast exocytosis were investigated in the experiments where DM-nitrophen was activated using a xenon flash lamp (13). The membrane capacitance was measured with a time resolution of 1 ms in these experiments. We detected a delay ranging between 4 and 20 ms in all 9 cells examined (an arrowhead in Fig. 1C). The onset of exocytosis was determined from the intercept with the baseline of the linear extrapolation of the early part of the exocytosis (13). The delays were estimated as the time between the onsets of the flash and exocytosis, and were Ca 2ϩ -dependent and shorter at higher [Ca 2ϩ ] i (Fig. 1F).
The Ca 2ϩ -dependent Endocytosis-The fast exocytosis was followed by decreases in the membrane capacitance in 28% of the cells (Fig. 2; 14 out of 49 cells). The amplitudes of the endocytosis ranged between 20 and 200% (92 Ϯ 43%, mean Ϯ S.D., n ϭ 14) of that of the exocytosis. The time constants of the endocytosis ranged between 2.5 and 4 s. Decreases in membrane capacitance were selectively induced often at the second Ca 2ϩ jump (Fig. 2B) as in the case of endocrine cells (11). The endocytosis often occurred with a delay of 0.5 to 2 s (arrowheads in Fig. 2). Unlike in the case of the large dense-core vesicle secretion in endocrine cells (22,23), the endocytosis was never accompanied by a stepwise decrease in capacitance.
The Slow Ca 2ϩ -dependent Exocytosis-Some large stepwise capacitance increases were detected in 40% of the cells where Ca 2ϩ jumps larger than 5 M were observed (Fig. 1C and Fig.  3B). The sizes of the steps ranged between 5 and 134 fF, representing the exocytosis of vesicles with diameters between 0.4 and 1.5 m. Notably, we found that the stepwise capacitance increases were selectively induced at low [Ca 2ϩ ] i (Ͻ20 M) where little fast exocytosis was detected (n ϭ 9, Fig. 3A), indicating that this component of exocytosis has a higher affinity for Ca 2ϩ than the fast exocytosis. The minimum [Ca 2ϩ ] at which the stepwise increases were observed was 6 M. Precise Ca 2ϩ dependence of the stepwise exocytosis, however, was difficult to confirm, because the steps occurred infrequently. The numbers of steps (Ͼ10 fF) induced by a Ca 2ϩ jump ranged between 1 and 10 (mean Ϯ S.D. ϭ 3.6 Ϯ 2.6, n ϭ 24).
The exocytosis of the stepwise capacitance increase was slower than the fast exocytosis. We estimated the speed of the stepwise exocytosis in those cells in which Ca 2ϩ jumps larger than 30 M were evoked. The cumulative latency histogram of the large steps demonstrated that the rate constant of the slow exocytosis was about 0.3/s, about one-tenth that of the fast exocytosis. DISCUSSION We have found that the CHO fibroblasts display massive capacitance increases upon Ca 2ϩ jumps, and identified two types of capacitance changes. The fast component of capacitance increases occurred smoothly and showed a lower affinity for Ca 2ϩ . In contrast, the slow capacitance increases occurred in a stepwise manner and exhibited a higher affinity for Ca 2ϩ . We will discuss below the possibility that these two types of capacitance changes reflect secretion via two distinct exocytotic pathways.
The fast components of capacitance changes of CHO cells showed many properties in common with the exocytosis reported in nerve terminals and endocrine cells. First, the capacitance changes exhibited an exponential time course and occurred with a delay (Fig. 1, B and C) (10,12,13,24). Second, the capacitance changes could be exhausted by large Ca 2ϩ jumps (Fig. 2B) (11,23). Third, the capacitance changes showed a rather low sensitivity to Ca 2ϩ , and depend on Ca 2ϩ in a cooperative manner (Fig. 1E) (11-13). Finally, the capacitance increases were followed by capacitance decreases, possibly reflecting endocytosis, which also occurred with a delay (Fig. 2) (12,23,24). Thus, capacitance changes detected in CHO cells were phenomenologically very similar to those of secretory cells. The rate constant of the fast exocytosis in CHO cells, however, was one-tenth that of endocrine cells and one-thousandth that of neurons. These data were consistent with the hypothesis that CHO cells possess a Ca 2ϩ -dependent exocytosis pathway akin to those in secretory cells, although its rate constant was far smaller. It can be speculated that the fast exocytosis of CHO cells more likely reflects exocytosis of the small synaptic-like microvesicles, because the endocytosis following the fast exocytosis was not associated with large stepwise decreases in the membrane capacitance, reflecting vacuolation, as in the case of secretion via the dense-core vesicles in endocrine cells (12,22,23).
In order to compare the kinetic features of the Ca 2ϩ -dependent exocytosis in CHO cells and those in neuron, their rate constants and delays were fitted to a model used to explain those of the exocytosis in a nerve terminal (13): where V0 to V4 represent vesicles whose Ca 2ϩ binding sites are occupied by 0 to 4 Ca 2ϩ , F a fused vesicle, k the rate constant of Ca 2ϩ binding, C the concentration of Ca 2ϩ , l the rate constant of Ca 2ϩ dissociation, and a the rate constant of fusion. A parameter c is to account for the cooperativity of the Ca 2ϩ binding (13), and is assumed as 0.2. We estimated the rate constant of fusion, a, as 3/s from the maximum rate of exocytosis at high [Ca 2ϩ ] i . Then, the delay in the exocytosis was chiefly attributed to k, and the Ca 2ϩ sensitivity to k and l. A satisfactory fit of the data was obtained by setting k ϭ 7 ϫ 10 6 /s/M and l ϭ 2300/s (the solid line in Fig. 1, E and F). Thus, the fast exocytosis in CHO cells can be accounted for by a thousand times smaller fusion rate (a) and a slightly smaller association rate constant (k) than those of the nerve terminal. Thus, the major characteristic of exocytosis in CHO cells appears to be due mainly to slow fusion machinery downstream of Ca 2ϩ binding. The lower rate may be due to subtypes of synaptotagmin (7) or proteins involved in the SNARE complex (3). Morimoto et al. (9) measured the mean latency (delay of onset of EPSC-like events) in the secretion of ACh from CHO cells as 3.6 ms when the cells were depolarized to ϩ50 mV by a constant current injection lasting 5 ms. We do not know the precise time course of [Ca 2ϩ ] i changes sensed by secretory vesicles in such experiment. However, even if we assume a very short Ca 2ϩ transient that starts at the onset of depolarization and that decays as fast as 1 ms, the above model predicted the mean latency as 13 ms at any peak [Ca 2ϩ ] i larger than 10 M. This latency should not be confused with the delay in the fast exocytosis evoked by sustained rises in [Ca 2ϩ ] i as shown in Fig.  1. Instead, the latency represents the time lag in exocytosis after transient rises in [Ca 2ϩ ] i are applied. The mean latency (13 ms) is calculated from the latency distribution function, dF(t)/dt, where F(t) represents a fraction of fused vesicle at a given time, t(s), in Equation 1, assuming changes in [Ca 2ϩ ] i as 100 exp(Ϫt/0.001) (M). The mean latency is almost insensitive to peak [Ca 2ϩ ] i and to the fusion rate, but they depend on the time course of [Ca 2ϩ ] i decay and the dissociation rate of Ca 2ϩ , l. The mean latency was not shortened, even when we assumed faster decay in [Ca 2ϩ ] i , such as 0.1 ms. Therefore, the difference in the actual (3.6 ms) and predicted (13 ms) latencies may be ascribed to the following factors. First, in the experiments done by Morimoto et al. (9), ACh may have been selectively loaded into a subpopulation of vesicles that had a larger Ca 2ϩ dissociation rate. Second, our model (Equation 1) may be imprecise, and, for example, the dissociation rate of Ca 2ϩ is erroneously estimated. Finally, there may be clonal variations among CHO cell lines. We are planning to resolve this issue by combined ACh detection and photolysis of caged-Ca 2ϩ compound in our CHO cell line.
We have found that CHO cells displayed stepwise increases in membrane capacitance, indicating the secretion of large secretory vesicles with diameters of between 0.4 and 1.4 m. The large vesicle secretion was slower and showed a higher affinity to Ca 2ϩ than the fast exocytosis (Fig. 3). Thus, the slow exocytosis in CHO cells is similar to the exocytosis of the dense-core vesicles in nerve terminals in that it was slower, more sensitive to Ca 2ϩ , and mediated by larger secretory vesicles than the fast exocytosis (2,12). The large vesicle exocytosis in CHO cells is, however, very infrequent compared with that of endocrine cells; none was detected in 60% of the cells, and, on average, 2.5 such exocytoses were detected.
Finally, the reason why the cell line possesses an abundance of Ca 2ϩ -dependent exocytotic pathways should be considered. If such pathways were not necessary for survival or proliferation of the cell line, they might not have been evolutionary preserved. They may traffic membrane proteins or lipids to the plasma membrane or secrete extracellular matrix proteins. Ca 2ϩ -dependent incorporation of a membrane integral protein, NCAM, has been reported in neurons and endocrine cells (25). The Ca 2ϩ -dependent exocytotic pathways might function also as constitutive secretory pathways, and not necessarily be coupled with Ca 2ϩ rises, as in the case of secretion via synaptic vesicles. In fact, spontaneous secretion of ACh was detected in CHO cells (9). Electron microscopic identification of secretory vesicles responsible for the two components of capacitance changes in CHO cells would facilitate understanding of Ca 2ϩregulated exocytosis.